How Leopoldo Keeps Your Plugins Up to Date

If you have spent any time building custom prompts or workflows for Claude Code, you know the pain. You craft something that works, share it with your team, and within a week there are five different versions floating around. Someone tweaked the system prompt. Someone else added a step but forgot to tell anyone. The original author made improvements that never reached the rest of the team.

This is the reality of manual prompt management: version confusion, inconsistent results, and constant maintenance overhead. It does not scale.

How Leopoldo's Update System Works

Every Leopoldo plugin ships with a built-in update engine. Updates are explicit and user-initiated. Nothing happens in the background without your knowledge.

1. Integrity Check on Session Start (Offline)

When your session begins, the update engine runs a lightweight integrity check. This is completely offline: it verifies that your installed files match the manifest, detects missing or corrupted files, and confirms your installation is intact. No network calls, no version server, no data sent anywhere.

2. Explicit Updates via /leopoldo update

When you want to check for new versions, run `/leopoldo update`. The engine authenticates with Leopoldo's backend API, compares your local version against the latest release, and downloads any available updates. You see exactly what changed before anything is installed.

3. Manifest-Aware Updates

The update engine respects your modifications. If you have edited a managed skill, that skill is marked as modified and updates are skipped for it. Your customizations are never overwritten. Only managed skills with unchanged hashes receive updates.

4. Snapshots Before Changes

Before applying any update, the engine creates a snapshot of your current installation. If anything goes wrong, you can roll back with `/leopoldo rollback`. Up to three snapshots are retained automatically.

Why This Matters More Than You Think

Security Patches Delivered on Demand

When a vulnerability is discovered and patched, the fix is available the next time you run `/leopoldo update`. One command and you are current.

Methodology Improvements Ship Continuously

Leopoldo's plugins encode structured methodologies: not just prompts, but complete systems of agents, orchestrators, and hooks working together. When we improve a methodology based on real-world usage data, that improvement is available to every user on their next update.

Team Consistency Without Team Effort

Every person on your team gets the exact same version of every plugin. No drift, no "which version are you on?" conversations, no shared docs with outdated instructions.

The Manual Alternative

Consider what maintaining equivalent tooling manually looks like:

• Copy-paste prompts from a shared doc or repo
• Remember to check for updates (you will not)
• Manually diff changes when someone updates the source
• Hope everyone on the team is using the same version
• No security scanning of prompt changes
• No structured versioning or rollback capability

This approach works for simple, personal prompts. It falls apart the moment you need reliability, consistency, or scale.

What Gets Updated

Updates cover the complete plugin kit: agents that orchestrate complex workflows, hooks that enforce quality gates, and the full library of domain expertise that makes each plugin a specialist. When any component improves, the entire kit updates as a unit, tested and verified before it ever reaches your machine.

Built for Professionals Who Ship

The update system reflects a core belief: professional tools should be easy to maintain. You would not accept a security tool that required manual signature updates. You would not use a package manager that did not resolve dependencies. Your Claude Code plugins should be just as straightforward to keep current.

Explore all domains at leopoldo.ai/domains. Full Stack is free on GitHub. Specialized domains available on request at hello@leopoldo.ai.

Want to see these plugins in action? Browse the catalog.